Glucose meters are widely used in hospitals for point-of-care monitoring of blood glucose. However, concerns about their accuracy and reliability in a neonatal setting have been expressed. We evaluated the analytical performance and clinical reliability of a new-generation glucose monitoring system (StatStrip; Nova Biomedical, Waltham, Mass) that corrects for interference factors by comparing results with a laboratory plasma hexokinase reference method, along with 3 other glucose meter systems.Methods:
Analytical studies for assessing precision and method correlation in the laboratory were performed comparing glucose to plasma glucose results from a reference hexokinase method. Challenge studies were performed with interfering substances including hematocrit, maltose, ascorbate, and β-hydroxybutyrate (βHB) that have previously been shown to affect measurements from current glucose meter technologies. Clinical reliability was assessed with 109 capillary blood specimens collected from neonatal intensive care unit (NICU) patients with varying hematocrit levels.Results:
Within-run imprecision was similar for all 4 meter systems, with the coefficient of variation generally less than 5%. Of the 4 meter systems, 2 (StatStrip and Contour [Bayer Healthcare Diabetes Care, Newbury, UK]) showed closer correlation to the reference hexokinase method after regression analyses on 75 spiked whole blood specimens. Linear regression analysis demonstrated a regression line slope of 0.960 for Contour and 0.920 for StatStrip, with lower slope values of 0.705 and 0.791 for Optium (Abbott Diabetes, Alameda, Calif) and Advantage (Roche Diagnostics, Indianapolis, Ind), respectively. The Contour and StatStrip glucose meter systems had the lowest mean biases compared with the hexokinase method. Of the 4 strip meter systems tested for interference, only the StatStrip and Optium systems remained within 0.55 mmol/L of their initial value (at a glucose concentration <5.55 mmol/L) and 10% (at a glucose concentration >5.55 mmol/L) after the addition of ascorbate, maltose, or βHB. Maltose had a marked effect on the Advantage system, whereas ascorbate and βHB had only minor effects on the Contour system. Increasing hematocrit values significantly lowered the glucose value readings for samples with medium to high glucose values tested by the Optium and Advantage systems. Varying hematocrit levels had minor and clinically insignificant effects on both the StatStrip and Contour systems. This was further substantiated after bias plot analysis of hematocrit levels in the NICU patient population. When applied to NICU patient samples, StatStrip showed close accordance with ISO 15197 performance criteria and good accuracy in comparison with the laboratory hexokinase method.Conclusions:
Managing glucose levels in a neonatal critical care setting is dependent on using reliable and accurate glucose meters with good specificity. The StatStrip and Contour meters performed well based on evaluation, with no interference observed for hematocrit, maltose, βHB, and ascorbate. In the clinical study on the NICU patients, the StatStrip meter showed good correlation with the hexokinase method across the analytical range.